arX
iv:a
stro
-ph/
9704
037v
1 3
Apr
199
7
Prototype Test Results of the Solar Tower Atmospheric Cherenkov Effect
Experiment (STACEE)
M.C. Chantell1, D. Bhattacharya2, C.E. Covault1, M. Dragovan1, R. Fernholz3, D.T. Gregorich4,5,
D.S. Hanna3, G.H. Marion1, R.A. Ong1, S. Oser1, T.O. Tumer2, D.A. Williams6
1 University of Chicago, Chicago IL 60637, USA
2 University of California, Riverside, Riverside, CA 92521, USA
3 McGill University, Montreal, Quebec H3A 2T8, Canada
4 California State University, Los Angeles, Los Angeles, CA 90032, USA
5 IPAC, California Institute of Technology, Pasadena, CA 91125, USA
6 University of California, Santa Cruz, Santa Cruz, CA 95064, USA
(Submitted to Nuclear Instruments and Methods in Physics Research A)
Abstract
There are currently no experiments, either satellite or ground-based, that are sensitive to astro-
physical γ-rays in the energy range between 20 and 250 GeV. We are developing the Solar Tower
Atmospheric Cherenkov Effect Experiment (STACEE) to explore this energy range. STACEE will
use heliostat mirrors at a solar research facility to collect Cherenkov light in extensive air showers
produced by high energy γ-rays. Here we report on the results of on-site test work at the solar
facility. We demonstrate that the facility is suitable for use as an astrophysical observatory, and
using a full scale prototype of part of STACEE, we detect atmospheric Cherenkov radiation at en-
ergies lower than any other experiment to date. Based upon these results we are confident that the
eventual STACEE instrument will be capable of exploring the γ-ray sky between 50 and 500 GeV
with good sensitivity.
PACS codes: 95.55.Ka, 07.85.-m, 42.79.Ek 29.40.Ka
1
http://arxiv.org/abs/astro-ph/9704037v1
1 Introduction
During the last few years the field of γ-ray astronomy has been revolutionized by the discovery of over
100 point sources by the EGRET satellite experiment with energies up to 20 GeV [1]. At the same time,
improvements in ground-based telescopes using the atmospheric Cherenkov technique have resulted in
several recent detections of point sources at energies above 250 GeV [2]. Still, most EGRET sources are
not detected above 250 GeV. For example, of the large number of Active Galactic Nuclei (AGN) seen by
EGRET, only Markarian 421 has been detected by a ground-based instrument [3]. This result implies
a spectral cutoff between 20 and 250 GeV, and may suggest that high energy γ-rays are attenuated by
photon-photon interactions with the intergalactic infrared background [4, 5]. A measurement of AGN
spectral cutoffs as a function of AGN redshift may be used to probe the infrared background, which is
sensitive to details of galaxy formation and dark matter composition [6]. As another example, although
it is generally believed that supernova remnants (SNR) are important sites for cosmic ray acceleration,
to date no clear detection of γ-rays from SNR has been made by ground-based instruments [7]. There
is speculation that the γ-ray spectra of SNR soften at energies above 20 GeV [8], making it important
to observe such objects at as low an energy as possible. For these reasons, the energy range between 20
and 250 GeV is expected to yield a wealth of scientific discovery. To date, however, it remains largely
unexplored. The existing satellite experiment (EGRET) has poor sensitivity above ∼ 10 GeV due to its
limited collection area, while current ground-based experiments have energy thresholds above 250 GeV.
When high energy γ-rays or cosmic rays enter the Earth’s atmosphere, they interact and produce ex-
tensive air showers (EAS) of highly relativistic charged particles. These charged particles emit Cherenkov
radiation which forms a light pool ∼ 100m in radius at ground level. Atmospheric Cherenkov telescopes
operate by using large mirrors to collect this light and focus it onto photomultiplier tube cameras. The
total amount of Cherenkov light generated by an EAS is directly proportional to the energy of the
progenitor and thus, as one goes down in energy, the density of Cherenkov photons on the ground de-
creases. The energy threshold of this type of instrument is limited by the total mirror collection area.
Larger mirror areas yield lower energy thresholds. Currently, the lowest energy threshold obtained by
an atmospheric Cherenkov telescope is ∼ 250 GeV for the Whipple Observatory’s 10m (78.5m2 mirror
area) telescope.
It has been recognized for some time that existing solar power plants represent a potential resource
for achieving lower energy thresholds due to their large mirror areas [9, 10]. Over the last few years,
we have been exploring the use of heliostat fields at such facilities, with the goal of developing a new
experiment called the Solar Tower Atmospheric Cherenkov Effect Experiment (STACEE) [11, 12, 13].
2
The goal of STACEE is to explore the γ-ray sky in the energy range of 50 GeV to 500 GeV. Our current
plan is to make use of the National Solar Thermal Test Facility (NSTTF) located in Albuquerque, NM.
STACEE will use 48 heliostats, spread out over a 150m × 300m area, to sample a large fraction of the
Cherenkov light pool. Each heliostat has a mirror area of 37.2m2, yielding a total mirror collection area
of ∼ 1786m2.
In 1996, the STACEE collaboration designed and built a prototype secondary telescope and camera
for detecting atmospheric Cherenkov showers using the heliostat field of the NSTTF This prototype was
installed on the central receiver tower at the NSTTF for a series of tests in August and October 1996.
The prototype included electronics and a data readout system for eight photomultiplier tubes viewing
eight heliostats.
The purpose of these tests was to explore the feasibility of establishing a γ-ray observatory at the
Sandia site. In particular, our goals for these tests were to establish the following:
• that the environmental conditions at the Sandia site are suitable for doing atmospheric Cherenkov
astronomy,
• that the mechanical and optical properties of the heliostats are of sufficient quality for astrophysical
observations, and
• that the STACEE concept is viable for a low threshold ground-based γ-ray detector.
Our results on all three of these topics were excellent. In the remainder of this paper we elaborate
our findings.
2 Detector Concept
The STACEE detector uses large steerable mirrors, called heliostats, to collect Cherenkov light from
extensive air showers. This light is reflected onto a secondary mirror mounted on the central receiver
tower. The secondary mirror images the light onto an array of photomultiplier tubes (PMTs) which are
mounted on a supporting structure. Because the secondary mirror forms an image of the heliostat field
in its focal plane, each PMT may be positioned so that it sees light from only one heliostat. The PMTs
together with the supporting structure are referred to as the PMT camera. Figure 2.1 shows a schematic
of the STACEE concept. Complete details of the STACEE design concept can be found in [14].
In our 1996 tests the signals from each PMT were capacitively coupled, amplified, and delayed to
correct for the varying arrival time of the Cherenkov light at the camera box. PMT signals were combined
3
to form a fast trigger. Analog to digital converters (ADCs), time to digital converters (TDCs), and
waveform digitizers recorded the pulse amplitudes, times, and waveforms, respectively. The data were
read out via CAMAC and GPIB by a 486-based PC. The prototype tests used eight heliostats with
eight PMTs. Figure 2.2 shows a schematic of the secondary telescope and camera and Figure 2.3 shows
a schematic of the electronics setup. Different sets of heliostats were used in the August and October
1996 tests, as indicated in Figure 2.4.
3 Results from Environmental Measurements
The NSTTF is located on the grounds of Kirtland Air Force Base near Albuquerque, New Mexico, at
an altitude of 1700m. Environmental conditions relevant to the STACEE project include:
• the local weather conditions, particularly the expected number of clear nights for astronomical
observing,
• the clarity of the atmosphere, particularly the impact of air pollution from nearby Albuquerque
that might attenuate Cherenkov light, and
• the ambient light levels at night – including the impact of any extraneous light from Albuquerque.
3.1 Local Weather Conditions
Sandia National Laboratories are situated in the southwestern United States in an area with a dry
moderate climate. With the exception of seasonal monsoons from July through August, this region
receives little precipitation. Atmospheric levels of water vapor, which increases the attenuation of light,
are low, and the skies are usually very clear. From meteorological records for Albuquerque (Figure 3.1)
we estimate that we will achieve an average of 4.1 hours of cloudless, moonless weather per night during
the Sept. through May observing season. We estimate an annual duty cycle of ∼ 10%, which is typical
for ground-based atmospheric Cherenkov experiments at good observing sites.
3.2 Atmospheric Clarity
Atmospheric contaminants, such as aerosols and other pollutants, have the potential to attenuate the
Cherenkov light signal as it propagates through the atmosphere above the Sandia site. To determine the
clarity of the local atmosphere at Sandia we used stellar photometry to measure the optical transmission
of the atmosphere at the site as a function of atmospheric depth and wavelength.
The photometry data were collected over the course of a single night at Sandia at a location ap-
proximately 20 meters north of the north edge of the heliostat field. The photometer, as depicted
4
schematically in Figure 3.2, consists of a Hamamatsu photon-counting photomultiplier tube (H3460-53)
attached to the focus of a Celestron 8" F/10 Schmidt Cassegrain Telescope (SCT). The PMT base
housing contains amplifier and discriminator circuitry. Output pulses from the PMT are counted with
a scaler which is read out by a laptop PC. The count rate from the PMT is directly proportional to
the photon flux falling on its photocathode. The instrument includes a set of standard Johnson UBV
photometric filters that allows measurements to be made in the Ultra-violet, Blue, and Visible portions
of the spectrum respectively. Figure 3.3 shows the standard UBV photometric response curves. We
note that the corrector plate of the SCT is made of crown glass and hence does not transmit below
wavelengths of about 340 nm. This cutoff affects measurements in the U band, and is corrected for in
the subsequent analysis of the data.
We made repeated drift scans of four separate bright stars over a five hour period. Drift scans were
done for each star using each filter. Each star was scanned at five different zenith angles. The scans
included approximately 40 seconds of data before star transit in order to measure the background light
level. Figure 3.4 shows the data from a single scan. The background and signal regions are clearly
separated, and the boundaries of the signal region are very sharp and well defined.
The total atmospheric transmission factor for each of the UBV wavebands is derived from the ob-
servations using standard photometric techniques [15]. The observed flux of star light is found from the
data by calculating the average background rate from the data before the the signal peak (see Figure
3.4) and subtracting that value from the average count rate in the signal region of the drift scan, taking
care to avoid the regions where the signal is turning on. Error bars shown on subsequent plots are the
statistical errors associated with these averages.
To find the unattenuated flux of star light we use Bouguer’s law [15] which relates the apparent
magnitude of a star (Mo,λ) at the zenith as a function of wavelength to (Mz,λ) the magnitude as a
function of zenith angle z and wavelength:
Mz,λ = Aλ · sec z +Mo,λ. (1)
Note that the logarithm of the photon flux is proportional to the stellar magnitude M . To first order,
sec z is proportional to the atmospheric depth (sometimes called the airmass or overburden). From a
plot of Mz,λ versus sec z the unattenuated stellar magnutide can be found by extrapolating to sec z = 0
which corresponds to an airmass of 0 (ie. no atmosphere). Figure 3.5 shows such a plot of the data for
the star Mu Andromeda with the V band filter.
5
The atmospheric transmission is calculated as the ratio of the stellar flux on the ground to the
incident stellar flux at the top of the atmosphere. A weighted average of results from all four observed
stars yields the total atmospheric attenuation for each wavelength band.
Figure 3.6 shows the atmospheric transmission derived from the observations for the blue waveband
as a function of zenith angle. The solid line shows the fit of the data to the form:
T = a exp(−b sec(z)), (2)
where T is the atmospheric transmission factor, a and b are free parameters. Table 1 gives the atmo-
spheric transmission at zenith for all three wavebands.
Figure 3.6 also shows the total transmission predicted from a model of an ideal atmosphere. This
model assumes that the only light loss mechanisms are Rayleigh and Mie scattering and absorption
by oxygen and naturally-occurring ozone [16]. Instrumental effects such as PMT quantum efficiency,
wavelength dependent transmission of the crown glass corrector plate, the reflectivity of the mirrors in
the telescope, and the appropriate transmission for the UBV filters have all been folded in to the model.
We can see that the ideal atmosphere agrees with the observational data to within 3% (see Table 1).
These measurements thereby demonstrate that the clarity of the local atmosphere is not significantly
affected by air pollution. Table 1 also shows the predicted clarity for the Whipple Observatory for
comparison, calculated assuming an ideal atmosphere; the Sandia and Whipple sites have comparable
atmospheric clarity.
3.3 Ambient Background Light
In addition to Cherenkov light from extensive air showers, the STACEE detector is also sensitive to
ambient light present at the site. This ambient light constitutes the background against which the
Cherenkov signal must be detected. One source of ambient background is light from the night sky viewed
by the heliostats. This includes air glow, stars in the field-of-view of the heliostats, and backscattered
light from artificial sources on the ground. Because the secondary mirror looks down into the heliostat
field, a second source of ambient light is light scattering off the ground surrounding the heliostats. This
background can be minimized by matching the aperture of each PMT to the size of the heliostat image
in the focal plane, so that the PMT sees only the surface of the heliostat and not the ground around it.
Figure 3.7 illustrates the relative contributions of these backgrounds when the heliostats are placed
in different orientations. The largest background is seen when the heliostats are pointed to reflect light
arriving from zenith into the STACEE telescope. Since we use Winston cones to limit the field of view
6
of each PMT to the area of a single heliostat, we expect very little of the reflected light from the ground
to be picked up by the PMTs when the heliostats are pointed towards the sky. When the heliostats
are turned so that they are “edge on” as seen by the STACEE telescope, the PMTs view only the light
reflecting off the ground around the heliostats, and the level of the background drops by a factor of ∼ 1.5.
Since the background light level is reduced by taking the heliostats off the night sky we are confident
that the ambient light entering our PMTs is dominated by light from the night sky and not by light
produced by nearby Albuquerque.
We calculate the flux of background photons from the zenith by measuring the currents in the PMTs.
The current from each PMT due to background can be expressed as:
I = Φbkg e G Ω ǫ A, (3)
where I is the PMT current in amperes, Φbkg is the background photon flux, e is the charge of the
electron in Coulombs, G is the PMT gain, Ω is the solid angle in the sky viewed by each heliostat, ǫ is
the efficiency with which a photon incident on the heliostat results in a photoelectron at the PMT, and
A is the heliostat collection area. We find that the flux of photons is 4.3± 0.9× 1012 ph/m2/sec/sr. For
comparison, the measured flux at a dark mountain site is 2.0× 1012 ph/m2/sec/sr [17].
4 Results from Heliostat Performance Tests
The heliostat field at Sandia is designed to track the Sun and focus its light onto the central receiver
tower. An individual heliostat consists of 25 square mirror facets, each 1.22m on a side, mounted on a
single altitude-azimuth mount. The surface of each facet is deformed slightly using adjustable screws to
obtain a parabolic reflective surface. The facets on each heliostat are aimed and focused to produce the
smallest possible image of the Sun on the receiver tower.
The elevation and azimuth positions of each heliostat are encoded with a precision of 13 bits over
360◦ which corresponds to a precision of 0.04◦. Pointing, tracking, and all other aspects of heliostat
operations are implemented by a central controller.
To determine the suitability of the existing heliostats at the Sandia site for astronomical measure-
ments, we measured the following:
• pointing accuracy,
• tracking stability,
7
• focusing properties of the heliostats, and
• reflectivity.
Details for each of these measurements are described below.
4.1 Heliostat Pointing Accuracy
Heliostat pointing accuracy is important to ensure that: 1) all heliostats are viewing the same point in
the sky, and 2) the collected light is properly focused onto the secondary optic on the tower.
The pointing accuracy of the heliostats was determined by conducting drift scans of several bright
stars. During a single observation, all heliostats were directed to observe a point seven minutes in right
ascension ahead of a bright star. The heliostats were halted and the star was allowed to drift through
the field of view of the heliostats while the PMT currents were recorded with a scanning ADC. Figure
4.1 shows the light curves obtained for two heliostats during a drift scan of the star Aldebaran. The time
of the peak current for each PMT is found from the weighted average of the background-subtracted light
curve. Table 2 lists the time, in seconds from the start of the data run, at which the current reached a
maximum in each heliostat’s light curve.
The data show that the selected heliostats were all aimed at a common point to an accuracy better
than 0.08◦ with a typical accuracy of 0.04◦ which is equal to the resolution of the 13 bit heliostat position
encoders. Thus using star transit data we can readily identify heliostats which are not properly aligned
and make corrections to fine-tune the heliostat pointing.
The light curves obtained from these drift scans can also be used to determine the field of view for
each heliostat. We define the field of view of a heliostat to be the full width half maximum (FWHM) of
its recorded light curve for a star transit. From the known angular velocity of Aldebaran, we convert the
FWHM of each light curve into a field of view in degrees. For the two light curves shown in Figure 4.1,
we obtain an average heliostat field of view of 0.7◦.
4.2 Tracking Stability
The ability of a heliostat to maintain a celestial object centered in its field of view as the object moves
across the sky is referred to as the tracking stability. To examine the stability of the Sandia heliostats,
we tracked the bright star Betelgeuse for 18 minutes while recording the PMT currents with scanning
ADCs. Figure 4.2 shows the light curves for two of the heliostats. The average current level for each
heliostat is stable to within a few percent over the duration of the data run. There is an obvious sawtooth
modulation of the currents with a period of ∼ 30 sec. The sawtooth pattern seen in the lower plot of
8
Figure 4.2 results from the heliostat repeatedly being brought on target (the point of maximum current
in the PMT) and then slowly drifting off target. The distance through which the target drifts before the
heliostat is commanded to update its position is equal to the single bit resolution of the heliostat position
encoders. Knowing the apparent motion of Betelgeuse across the sky, we find that the periodicity of the
observed sawtooth pattern corresponds precisely to the period expected from the encoder resoultion.
Examination of data from another heliostat shown in the upper plot of Figure 4.2 reveals the same
type of sawtooth pattern but with the orientation of the “teeth” in the opposite sense. In this case the
heliostat was tracking a position slightly ahead of the target star. Thus the current dropped suddenly
when the heliostat moved to reacquire the target and then slowly rose as the star drifted back into the
field of view of the heliostat. These results demonstrate how star tracking can be used as a powerful
diagnostic tool for evaluating the pointing bias of individual heliostats. We plan to conduct regular star
tracking runs as a means of monitoring the heliostat tracking stability. Through the use of these runs
and stellar drift scans, we expect to be able to correct the pointing biases of all 48 heliostats to within
0.04◦. This accuracy is more than adequate, given STACEE’s expected angular resolution of ∼ 0.2◦.
4.3 Optical Properties of the Heliostats
The critical optical property of the heliostats is their focusing. In order to maximize light collection it is
important that the reflected light from a heliostat be focused onto as small an area on the central tower
as possible. The minimum acceptable size for the secondary mirrors is determined by the projected
heliostat spot size.
The heliostat optics were evaluated by imaging the Sun onto the front face of the tower and recording
the images with a CCD camera. Fifteen of the sixteen heliostats used in the two prototype tests were
tested in this manner; we were unable to obtain data for one heliostat due to cloudy weather. Figure 4.3
shows a typical heliostat image of the Sun. From these images we have determined the average FWHM
heliostat spot size to be less than 2.0m diameter. Since the Sun has an angular extent of 0.5◦, the
size of its image closely matches the expected size of a 50 GeV γ-ray air shower. Table 3 lists the Sun
spot sizes for all fifteen heliostats tested. Figure 4.4 shows the FWHM of the spots as a function of the
distance between each heliostat and the target. These data indicate a regular trend towards a tighter
concentration of light for heliostats closer to the central tower. We can use this trend to predict the light
profile expected at the target from any heliostat in the field.
9
4.4 Heliostat Reflectivity
To estimate the total light collection efficiency of the STACEE optical system, it is necessary to know
the reflectivity of the heliostats as a function of wavelength. The heliostats at Sandia are back-surface
silvered glass approximately 3 mm thick. Because the glass is opaque to UV light, a significant amount
of the Cherenkov flux from air showers is unavoidably lost at the heliostat. Measurements performed
by Sandia personnel place the average heliostat reflectivity at 80% for visible light. To measure the
reflectivity as a function of wavelength, we designed and constructed a custom reflectometer which was
used in the field. The reflectometer uses a collimated light source and a high quality PMT to measure
light reflected from the surface of a mirror. The instrument is calibrated using a standard mirror of known
reflectivity. Narrow bandpass filters allow the measurement of reflectivity as a function of wavelength.
The results are shown in Figure 4.5. These measurements demonstrate that the overall reflectivity is
∼ 85%, and that the UV cutoff occurs near 330 nm. Since the heliostats are already over 20 years
old, and have maintained a high degree of reflectivity, we do not expect their optical performance to
significantly degrade over time scales relevant to the STACEE project.
5 Prototype Performance Results
In addition to establishing the suitability of the Sandia site for astronomical observations, we also built
and tested a fully functional STACEE prototype (See Figure 2.2). The prototype included a secondary
telescope, PMT camera, and data acquisition electronics for 8 channels. In both the August and October
tests, we ran the prototype STACEE experiment and investigated the following:
• optical characteristics of the secondary mirrors,
• performance of the analog trigger,
• performance of the digital trigger and coincident trigger rates,
• performance of waveform digitizers.
• energy threshold obtainable, and
The secondary telescope and camera were installed on one of the test bays of the central receiver tower.
This bay is 10 ′ deep, 35 ′ wide, and 160 ′ above the heliostat field, and looks out on the field to the
north.
10
5.1 Optical performance of the secondary mirrors
A critical component to the STACEE telescope is the secondary mirror that collects the Cherenkov
light from the heliostats and reflects it on to the PMTs. We have explored several design approaches
for developing large, highly reflective mirrors at low cost. Two distinct mirror technologies have been
tested in the field. During the August test we used a 3m diameter multifaceted secondary made with
a stretched aluminum membrane mirror technology. During the October test, we used a 1.8m diameter
back-silvered slumped glass mirror.
5.1.1 Stretched membrane mirrors
For the August test we used a secondary mirror with seven facets made of electro-polished stretched
aluminum sheet. The individual facets had spherical curvature and were aligned at night by directing a
high power searchlight beam onto a heliostat which reflected the light onto the secondary mirror. Each
facet produced an optical spot of approximately 4 cm diameter on a target at the focal plane. To co-align
the facets, six facets were covered with black cloth while the seventh facet was adjusted using a three
point turnbuckle arrangement. This procedure was repeated until the optical spots from all seven facets
were aligned. The aggregate alignment of the secondary system was cross-checked by tracking bright
celestial objects (such as the planet Jupiter) and projecting the collected light pattern onto a white lucite
board mounted at the focal plane.
5.1.2 Slumped glass secondary mirror
For the October test we used a slumped glass parabolic mirror (1.8m diameter) as the secondary optic.
Using a single mirror eliminates the need to align multiple facets. The slumped mirror also provided
superior optical surface quality, enough so that the images of the heliostats were clearly visible at the
focal plane during daylight. Figure 5.1 shows the images of eight selected heliostats projected onto a
white placard at the focal plane. Each of the eight heliostats is cleanly imaged by the secondary, with
virtually no optical overlap. The optical spot size for this mirror is less than 1 cm.
5.1.3 Optical crosstalk
For an ideal secondary mirror system, all of the reflected light from each heliostat in the field will be
focused onto the collecting aperture of a single PMT. However, optical aberrations, imperfections in
mirror surface quality, and facet misalignment may combine to defocus the image of the heliostats at
the focal plane, causing a small amount of light from one heliostat to enter another heliostat’s PMT.
We refer to this as optical crosstalk. We measure the amount of crosstalk using Cherenkov light signals
11
from cosmic ray air shower data. For each event, time-to-digital converters (TDCs) were used to record
the relative arrival times of the Cherenkov light pulses at each PMT. Since the time-of-flight depends
upon the unique distance between each heliostat and the secondary, photons that were reflected into the
wrong PMT will be out of time with other photons collected from the same Cherenkov event. Analysis
of the TDC differences between all combinations of adjacent channels show that the maximum level of
crosstalk is approximately 1%. There is no evidence for crosstalk between non-adjacent heliostats. The
largest amount of optical crosstalk occurs for physically adjacent channels in the same heliostat row.
(Note that with the application of cuts on TDC times and a strategic selection of heliostat locations,
optical crosstalk is expected to be negligible for the STACEE experiment).
5.1.4 Summary of Mirror Tests
Measurements made with two different mirror technologies show both are suitable for use as the secondary
optic. The image sizes are small enough to be contained by 5" diameter Winston cones. Optical crosstalk
is small and easily rejected off-line by exploiting the time-of-flight differences between different heliostats.
The single-piece slumped glass secondary gives an optically superior performance over the multi-
faceted aluminum secondary, and is also easier to mount and align. Therefore this technology is currently
considered the most promising for STACEE. The STACEE group continues to develop and improve
mirror performance using both technologies.
5.2 Analog vs. Digital Trigger
Atmospheric Cherenkov telescopes are typically triggered by requiring a minimum number of PMTs
to exceed a preset threshold within a small (∼ 10 ns) time interval. Two different methods exist for
implementing such a trigger. An analog trigger takes the analog signals from each PMT and sums them.
This summed signal is sent to a discriminator which produces a trigger if the summed signal exceeds a
preset threshold. In a digital trigger each channel is individually discriminated and a trigger is generated
when some minimum number of discriminated channels fire within a specified time interval. In deciding
on a suitable trigger scheme the following characteristics must be considered:
• sensitivity to accidental triggers resulting from fluctuations in the NSB and afterpulsing in the
PMTs,
• sensitivity to local phenomena such as single energetic muons passing close to an individual helio-
stat, and
• achievable energy threshold.
12
5.2.1 Analog Trigger
The primary advantage of an analog trigger is that it makes full use of the signal-to-noise improvement
that comes from increasing mirror collection area. It has been shown [18] that the achievable energy
threshold of an atmospheric Cherenkov telescope goes as:
Eth ∝
√
ΦbkgΩτ
ǫA. (4)
Here, Φbkg is the flux of night sky photons, Ω is the field of view, τ is the trigger gate width, A is
the mirror collection area, and ǫ is the efficiency for converting a photon striking a heliostat into a
photoelectron in its PMT. Of these quantities, Ω and τ are constrained by the physics of air showers and
ǫ is constrained by technology. The most direct way to lower the energy threshold of an atmospheric
Cherenkov telescope is to increase the effective mirror area. Since an analog trigger takes full advantage
of information contained in the time structure of the PMT pulses, in principle it provides the lowest
possible energy threshold.
We tested this form of trigger by pointing all eight heliostats to different parts of the sky, near
the zenith, so that their fields of view did not overlap. In this configuration, triggers result only from
unwanted sources of background, and not from Cherenkov radiation seen in common by a number of
heliostats. We measured the trigger rate as a function of discriminator threshold. The results are shown
in Figure 5.2. The data show a clear spectral break near a discriminator threshold of ∼ 200mV. If the
camera were triggering only on fluctuations of the background photon flux we would expect to see a
single, steeply falling spectrum. To understand the presence of the second spectral component at high
thresholds, we use Monte Carlo techniques, together with a model for the optical throughput of the
STACEE prototype, to estimate the expected trigger rates resulting from single energetic muons passing
near individual heliostats and for single heliostats triggering on small cosmic ray air showers.
To estimate the expected rates, we determine the number of photoelectrons necessary to cause a
trigger as a function of discriminator threshold from the known gain of the PMTs (∼ 0.6 × 106), the
average PMT pulse width (∼ 10 ns FWHM), and the measured losses due to cable attenuation (∼ 0.3).
The conversion between photoelectron signal and discriminator threshold is estimated to be 0.28 pho-
toelectrons/mV. Using the average photon density on the ground, as determined by Monte Carlo, we
estimate the energy threshold for protons and single muons as a function of discriminator threshold.
The trigger rate for a single heliostat can then be found from:
R = Φ Aeff Ω ǫ, (5)
13
where R is the trigger rate, Φ is the flux of Cherenkov-generating background particles (cosmic ray
primaries or muons), Aeff is the effective collection area for a single heliostat, Ω is the solid angle viewed
by a heliostat, and ǫ is the efficiency for converting photons into photoelectrons.
For cosmic ray air showers we use an effective all-particle cosmic ray flux which has been corrected
for the varying Cherenkov yield as a function of composition [12]:
Φcr = 9.1× 10−6(
E
1000 GeV)−1.67 showers/cm2/s/sr, (6)
where E is the effective energy threshold in GeV. For local muons we use the flux of muons above an
energy of 7 GeV [19]:
Φµ = 1.52× 10−3E−1.24 µ/cm2/s/sr, (7)
where E is taken to be 7 GeV which is approximately the energy below which Cherenkov production from
muons ceases. From drift scan data (see Section 4.1) we find that Ω for a single heliostat is∼ 1.2×10−4 sr.
From Monte Carlo simulations of proton showers and individual muons we determine Aeff to be 15400m2
for protons and 154m2 for muons. Since all eight heliostats are triggering independently on different
parts of the sky the total trigger rate for all eight heliostats combined is eight times the rate for a
single heliostat (cosmic ray triggers + µ triggers). The predicted rates are in good agreement with the
measured rates for pulses greater than 200 mV, as shown in Figure 5.2. This analysis demonstrates that
the simple analog trigger is sensitive to single heliostat triggers resulting from single muons and small
cosmic ray showers.
5.2.2 Digital Trigger
To test the rate of background triggers for a digital trigger we measured the coincidence rates for
different combinations of PMT threshold and multiplicity. As for the analog trigger we performed these
measurements with the heliostats pointed at different parts of the sky to determine the rate of unwanted
background triggers. Figure 5.3 shows the rate versus threshold curves for three different multiplicity
requirements. Note that in contrast to the analog trigger, here we do not see a break in the spectrum,
which indicates that the observed rates are due only to fluctuations in the NSB and not a secondary
source of unwanted triggers.
5.2.3 Comparison of Analog and Digital Triggers
From the background rate versus threshold curves for the two trigger types we see that the analog trigger
is sensitive not only to fluctuations of the background photon flux, but also to single heliostat triggers
14
from cosmic ray air showers and local muons. The digital trigger appears to be sensitive only to the
background photon flux. The analog trigger is sensitive to large amplitude signals in any individual
channel whereas the digital trigger demands that the light from an event be spread out over several
heliostats. Since we wish to trigger selectively on Cherenkov light from γ-ray initiated air showers,
which will uniformly illuminate heliostats over a 100m radius, a digital trigger or a modified analog
trigger that incorporates some multiplicity requirement is best suited to our needs.
5.3 Coincidence Event Studies
To fully understand the performance of the digital trigger, we studied the rate of coincident events when
the heliostats were aimed to a common point in the sky. Using these measurements, we:
• established that the system was triggering on cosmic ray showers,
• investigated the zenith angle dependence,
• investigated the rates as a function of heliostat canting angle, and
• determined the energy threshold for the prototype telescope.
5.3.1 Cosmic Ray Triggers
To establish that the STACEE prototype is triggering on genuine air shower events we took coincident
data with the heliostats in two separate viewing configurations: all heliostats viewing separate areas of
the sky near the zenith (random mode), and all heliostats simultaneously viewing the zenith (coincident
mode). A digital trigger condition of any four out of eight (4/8) PMTs at a PMT threshold of 31mV was
used, and PMT pulse height and timing information were recorded. When the heliostats were in random
mode we expected no coincidences due to cosmic-ray air showers. In this configuration we observed an
event rate of 0.2 Hz. When the heliostats were placed in coincident mode the event rate increased to 5.1
Hz. This indicates that our rate of genuine coincident events due to cosmic ray air showers was 4.9 Hz.
To further establish that the STACEE prototype was triggering on cosmic ray air showers we examine
the PMT pulse height distribution from the data taken with the heliostats viewing the zenith in coincident
mode. Since the field of view of each heliostat is well matched to the angular extent of an air shower,
the total light collected by a channel is directly related to the energy of the air shower progenitor.
We expect that the spectrum of observed pulse heights in the individual channels should match the
differential cosmic ray energy spectrum at 100 GeV. In this energy range, the cosmic ray spectrum has a
power-law dependence on energy, dN/dE ∝ Eα, where α is −2.65. We analyzed 35 minutes of data taken
15
with the heliostats observing the zenith in coincidence mode. The digitized pulse heights of all channels
exceeding threshold were combined to build a pulse height spectrum (Figure 5.4). The rising part of
the observed spectrum is due to a convolution of trigger inefficiencies for small pulses and fluctuations
in shower brightness at energies near threshold. The trigger becomes fully efficient for events with pulse
heights above 60 digital counts. A fit to the falling edge of the spectrum yields a power law index α
of −2.7 ± 0.1 (statistical error only), which agrees well with the known differential index of the cosmic
ray spectrum in this energy range. Thus, we are confident that we were in fact triggering on Cherenkov
light from cosmic ray air showers.
5.3.2 Zenith angle effects on system trigger
Because of the celestial motion of sources across the sky, γ-ray observations must be conducted over
a range of zenith angles. The effective energy threshold of an atmospheric Cherenkov telescope varies
with zenith angle due to the increased atmospheric overburden. As the overburden increases air showers
develop further away from the ground, resulting in: (1) an increase in the attenuation of the Cherenkov
light, and (2) a spread in the Cherenkov light pool over a larger area, reducing the photon density on the
ground. Both of these effects are proportional to the increase in atmospheric depth with zenith angle
and therefore, we expect the event rate to decrease with zenith angle. Figure 5.5 shows the observed
event rates as a function of zenith angle, which are well fit by a cos2(θ) function. Note that the event
rates fall off slowly out to a zenith angle of 30◦, after which they fall more steeply. In order to maximize
sensitivity, STACEE will concentrate on observing sources within 30◦ of the zenith.
5.3.3 Effect of heliostat canting on event rates
Since the heliostats have small fields of view, maximum trigger sensitivity is obtained when the heliostats
are pointed to the shower interaction region (approximately 10 km above sea level). The interaction
region is defined as the location where most of the Cherenkov light is generated for events that land on
the geometric center of the heliostat field. Tracking the heliostats to observe the interaction region, rather
than towards the source at infinity, requires a small adjustment in the pointing angle of each individual
heliostat. This adjustment is called “canting” because each heliostat is canted slightly inwards so as to
observe the interaction region.
Figure 5.6 illustrates the application of correct heliostat canting. Figure 5.6a shows the fields of view
of the heliostats in the case where all heliostats are tracking parallel in the direction of the source at
infinity. In this case, coverage of the event track is incomplete. Figure 5.6b shows the heliostats correctly
canted so that each heliostat will collect light from the full longitudinal development of the air shower.
16
The interaction region corresponds to the point of shower maximum in the development of an air shower
and occurs at an atmospheric depth of 270 g/cm2 (∼ 10 km altitude) for showers initiated by 50 GeV
γ-rays. Figure 5.7 shows the effect of changing the heliostat canting angle on the trigger rate. The
distribution is well fit by a Gaussian with a mean of 0.15◦, corresponding to a depth of shower maximum
consistent with the expected value for cosmic rays at energies of a few hundred GeV which compose the
bulk of the event triggers (see section 5.3.5).
5.3.4 Performance of Waveform Digitizers
A key goal of the on-site tests was to investigate the suitability of waveform digitizers for pulse timing
and amplitude measurement. As part of the electronics setup, we used Tektronix 644A and Tektronix
740A digital oscilloscopes to digitize the PMT waveforms for all eight phototubes. The Tektronix 644A
scope had a sampling rate of 2 GSample/sec, while the 740A had a sampling rate of 500 MSample/sec.
Sampling at 1 GSample/s was simulated with the 644A by averaging consecutive data points. Each
scope had a dynamic range of 8 bits. The digitized waveforms were read out over a GPIB interface and
saved to disk. The GPIB interface readout introduced a significant deadtime (nearly 0.4 sec per event),
and thus the waveform digitizers were only read out on selected runs.
For vertical showers, the maximum difference in shower arrival times at two adjacent heliostats spaced
10m apart is about 0.6 ns. (Note that here we refer only to shower arrival times at the heliostats in the
field and not the difference in shower arrival times at the secondary telescope on the tower). Thus, the
spread in photon arrival times is smaller than the expected measurement precision of ∼ 1 ns. Hence,
measurements of Cherenkov showers taken at zenith serve as a calibration beam for timing measurements,
and the spread in the difference of arrival times between two adjacent heliostats is dominated by the
timing resolution of the electronics.
The expected location of the Cherenkov pulse in each digitized waveform can be determined from
the known detector geometries and cable delays. Using data within a 15 ns window centered around the
expected pulse location, the arrival time of the Cherenkov pulse was defined to be the centroid of all
data points exceeding a threshold of 30 mV. We find that this technique yields better fitted times with
a smaller spread than other methods, including a Gaussian fit to the pulse shape.
To determine the timing resolution, we calculated the quantity:
∆t = t2 −t1 + t3
2, (8)
where, t1, t2, and t3 are the measured pulse arrival times at three consecutive heliostats in the same
row.
17
We calculate the RMS spread in ∆t, σ(∆t), for timing measurements made with the waveform
digitizers and for measurements made using conventional TDCs on the same data. Assuming independent
measurements, the timing resolution, σt, is equal to:
σt =σ(∆t)√
3
2
. (9)
We find that the waveform digitizers give a narrower spread in ∆t than the TDCs for all pulseheights
(See Table 4). In addition, the digitizers allow us to fit pulse arrival times for many pulses which were
too small to trigger the TDC. Thus, using the waveform digitizers, we are able to reconstruct arrival
times for events which cannot be reconstructed by TDCs alone.
We also studied the applicability of waveform digitizers for pulse charge measurement. By summing
the digitized voltages recorded within an integration region around the pulse arrival time, we determined
the total charge in the pulse. We find that the waveform digitizers give charge measurements which agree
with those found from conventional ADCs.
5.3.5 Energy Threshold
We estimate the effective energy threshold for cosmic rays and γ-rays of the eight heliostat configuration
used during the October test. We calculate the energy threshold for cosmic rays from the measured
event rate while observing the zenith. The observed rate is related to the energy threshold by:
R = Φcr Aeff Ω, (10)
where R is the observed rate, Φcr is the flux of cosmic ray showers, Aeff is the effective collection area,
and Ω is the solid angle acceptance of the instrument. For the flux of cosmic rays we use the equivalent
proton flux given by Equation 6.
The maximum observed trigger rate was 5.1 Hz, using a digital trigger with a multiplicity of 4/8
PMTs and a tube threshold of 31 mV. The accidental trigger rate in this configuration was 0.2 Hz, and
was subtracted from the total rate to yield 4.9 Hz.
The effective collection area and solid angle acceptance for proton-induced air showers are calculated
from Monte Carlo simulations. We simulate proton air showers using MOCCA [21], a well-established air
shower Monte Carlo. The simulated showers are then processed through a detailed ray tracing simulation
of the Sandia heliostat field. This simulation takes into account the measured focusing properties of the
heliostats and secondary mirror, reflectivity losses, and ambient background light levels. Proton showers
18
were simulated at energies from 50 GeV to 2 TeV and weighted according to the cosmic ray spectrum.
The simulated showers were then dropped randomly onto the heliostat field over a 100 m radius area
around the heliostats, and the incident angle of the showers were randomly varied within 1.0◦ of zenith.
The effective collection area calculated this way is:
Aeff = 1.9± 0.7× 108cm2, (11)
while the solid angle acceptance is:
Ω = 3.7± 0.2× 10−4sr. (12)
Note that the solid angle acceptance of the eight heliostats in coincidence is larger than the individual
heliostat acceptances. This is due to the fact that the system is triggering on air showers which have an
angular extent of 0.5◦.
Applying Equation 8 together with Equation 6 yields a cosmic ray energy threshold of:
Ecr = 295+96
−50 GeV. (13)
An alternative way of estimating the effective energy threshold is to compare the measured photo-
electron yields to the expected photon densities on the ground as a function of primary energy. We use
the MOCCA simulation to determine the average photon density on the ground within 100 m of the
shower core, as a function of energy (see Figure 5.8).
Figure 5.4 shows the pulse height spectrum built up from all PMT pulses which exceeded trigger
threshold for observations made at zenith. We define the trigger threshold pulse height as the pulse
height above which the integrated area of the power law fit equals the area under the data curve (ie.
the trigger rate predicted by the fit equals the observed rate). This corresponds to the minimum pulse
height needed to trigger a PMT. The trigger pulse height is found to be:
Ptrig = 33+8
−6 digital counts, (14)
where the uncertainty is due to the uncertainty in the spectral index of the fit. Knowing the rate at
which charge is converted to digital counts in the ADC (Q = 0.25 pC/digital count), the PMT gain
(Gpmt = 6.3 ± 1.3 × 105) and the gain of the pre-amp (Gamp = ×10) we can relate the equivalent
number of photoelectrons (Npe) at trigger threshold to the trigger threshold pulse height by:
19
Ptrig =Npe Gpmt Gamp e
Q, (15)
where e is the charge of an electron. From this we find:
Npe = 8.3+2.0−1.5 photoelectrons. (16)
To convert from photoelectrons at the photocathode to photons striking the heliostat (Nphot) we
divide by the average PMT quantum efficiency (0.21, a convolution of the Cherenkov spectrum on the
ground and the wavelength-dependent quantum efficiency of the PMT), and by the collection efficiencies
due to heliostat reflectivity (0.8), secondary mirror reflectivity (0.85), heliostat focusing efficiency on the
secondary (0.8), secondary focusing efficiency on the PMT winston cone (0.85), and the throughput of
the Winston cone (0.9). This gives the number of photons incident on a heliostat:
Nphot = 95+27
−23 photons, (17)
where a 15% systematic error has been included to account for uncertainties in the efficencies and
reflectivities.
Dividing by the projected area of a heliostat (29 m2) gives the photon density (ρ) on the ground
necessary to trigger a heliostat channel:
ρ = 3.3+0.9−0.8 photons/m
2. (18)
From Figure 5.8 we can use ρ to estimate the lowest energy γ-ray that will produce a high enough photon
density to trigger the experiment. We find the energy threshold, for γ-rays from a source at the zenith,
to be:
Eγ = 74+17
−14 GeV. (19)
For cosmic rays, the same photon density corresponds to an energy threshold of:
Ecr = 385+55
−60 GeV. (20)
This energy threshold agrees within error with that obtained from the rate calculation.
20
6 Summary
We are developing a novel atmospheric Cherenkov experiment called STACEE that will use solar heliostat
mirrors at the National Solar Thermal Test Facility (NSTTF). Two on-site tests have been conducted.
These tests have demonstrated that the heliostat field at the NSTTF is suitable for use as the primary
optical component of an astrophysical γ-ray detector. Weather and atmospheric conditions at the site
compare favorably to conditions at existing observatories. Measurements of the ambient background light
levels show that the close proximity of the site to Albuquerque will not adversely effect the performance
of STACEE. The mechanical and optical performance of the heliostat field is found to be excellent, in
most cases greatly exceeding our requirements.
The NSTTF is a scientific research facility, and so possesses an excellent support infrastructure. This
infrastructure includes heavy lift capability, a high bay, a complete machine shop, and other facilities
necessary for developing and running an experiment such as STACEE. We feel that the NSTTF site is
well suited in all respects for astrophysical observations.
The STACEE prototype’s secondary telescope, camera, and associated electronics also performed to
expectations. The secondary optics produce well-focused heliostat images that are completely contained
by the PMT Winston cones. Measurements of cosmic ray air showers show that the system is stable
and responds as expected. The measurements indicate that our prototype detector had a γ-ray energy
threshold below 100 GeV, and that the STACEE concept can successfully obtain lower energy thresholds
than existing ground-based techniques.
Acknowledgments
We are grateful to the Physics Division of Los Alamos National Laboratory for loans of electronics
equipment. We thank the staff at the NSTTF and in particular, acknowledge the contributions made by
J.M. Chavez, R.M. Edgar, C.M. Ghanbari, D. Johnson, J.J. Kelton, L. Killian, and R. Tucker. We also
acknowledge the assistance of E. Pod, the engineering staff of the Yerkes Observatory, the personnel of
the McGill Physics Department shop. We wish to thank M. Cresti for the use of a 1.8m mirror, and J.
Carlstrom of the University of Chicago for the loan of a digital oscilloscope. This work was supported
by the National Science Foundation, the Institute of Particle Physics of Canada, the Natural Sciences
and Engineering Research Council, and the California Space Institute. TOT wishes to acknowledge the
support of the University of California, Riverside, Vice Chancellor of Research and College of Natural
and Agricultural Sciences. CEC wishes to acknowledge support from the Louis Block Fund of the
University of Chicago. RAO wishes to acknowledge the support of the Grainger Foundation and the
21
Physical Sciences Division and the Enrico Fermi Institute of the University of Chicago.
22
References
[1] D. Thompson et al., Astrophys. J. Supp. 101, 259 (1995).
[2] R.C. Lamb et al., Proc. of the 1994 Snowmass Summer Study, Particle and Nuclear Astrophysics
and Cosmology in the Next Millennium, ed. E.W. Kolb and R.D. Peccei (World Scientific, Singapore)
295 (1995).
[3] M. Punch et al., Nature 358, 477 (1992).
[4] F.W. Stecker and O.C. De Jager, Astrophys. J. 415, L71 (1993).
[5] S.D. Biller et al, Astrophys. J. 445, 227 (1995).
[6] D. MacMinn and J.R. Primack, Space Sci. Rev. 75, 413 (1996).
[7] R.W. Lesard et al., Proc. 24th Int. Cosmic Ray Conf. (Rome) 2, 475 (1995).
[8] L. O’c. Drury, F.A. Aharonian, and H.J. Volk, Astron. Astrophys. 287, 959 (1994).
[9] S. Danaher et al., Solar Energy 28, 335 (1982).
[10] O.T. Tumer et al., Nucl. Phys. B (Proc. Suppl.) 14A, 351 (1990).
[11] O.T. Tumer et al., Proc. 22nd Int. Cosmic Ray Conf. (Dublin) 2, 635 (1991).
[12] R.A. Ong et al., Astroparticle Phys. 5, 353 (1996).
[13] R.A. Ong et al., Proc. Towards a Major Atmospheric Detector-IV (Padova, Italy), ed. M. Cresti,
261 (1995)
[14] The STACEE Collaboration, The Solar Tower Atmospheric Cherenkov Effect Experiment
(STACEE) Design Report, EFI Preprint 97-17, March 1997 (unpublished).
[15] C.R. Kitchin, Astrophysical Techniques (Adam Hilger, London), p. 304 (1991).
[16] M.C. Chantell, UV Imaging of Extensive Air Showers at TeV Energies, Ph.D. Dissertation, Uni-
versity of Arizona, 1995 (unpublished).
[17] R. Mirzoyan and E. Lorenz, Measurement of the Night Sky Light Background at La Palma,
MPI-PhE/94-35, (1994).
[18] T.C. Weekes, Phys. Reports 160, 1 (1988).
23
[19] O.C. Alkofer and P.K.F. Grieder, Physics Data - Cosmic Rays on Earth (Fach-Informations-
Zentrum, Karlsruhe) (1984).
[20] C.W. Allen, Astrophysical Quantities, 3rd Edition (The Athlone Press, London), p. 201 (1991).
[21] A.M. Hillas, Proc. 19th Int. Cosmic Ray Conf. (La Jolla) 3, 445 (1985).
24
Waveband U B V
Measured (Sandia) 0.548 0.707 0.813
Model Atmosphere (alt.=1700m) 0.562 0.726 0.841
Predicted Whipple (alt.=2300m) 0.590 0.760 0.861
Table 1: Atmospheric transmission factors for star light from the zenith. Note that 1700m corresponds
to the altitude of Sandia.
25
Heliostat ID Transit Time (s) ∆◦
8E3 433 0.05
8E4 433 0.05
8E5 402 -0.08
8E6 418 -0.01
10E4 437 0.07
10E5 421 0.00
10E6 412 -0.04
10E7 413 -0.03
Table 2: Star transit times for each heliostat for the drift scan of Aldebaran. The angular separation
(∆◦) of each heliostat, relative to the average of the eight transit times, is shown.
26
Distance Radius (m) Luminance within
Helio to target FWHM to contain luminance 2.0 m diameter
ID (m) (m) 50% 80% 90% (percent)
8E3 103.9 0.93 0.40 0.63 0.82 94
8E4 106.9 0.94 0.41 0.64 0.83 94
8E5 110.4 1.27 0.61 0.97 1.25 82
8E6 114.6 0.99 0.44 0.68 0.85 94
10E4 131.4 1.04 0.45 0.70 0.87 94
10E5 134.3 1.10 0.51 0.80 0.99 90
10E6 137.8 1.08 0.50 0.79 0.99 90
10E7 141.9 1.14 0.53 0.84 1.05 88
12W1 158.8 1.21 0.57 0.88 1.10 86
12E1 158.8 1.29 0.63 1.01 1.33 80
12E2 159.4 1.24 0.61 1.04 1.50 78
14E1 198.2 1.67 0.81 1.25 1.53 65
14E2 198.6 1.56 0.70 1.06 1.29 76
14E3 199.6 1.53 0.69 1.06 1.29 77
14E4 201.0 1.63 0.77 1.18 1.45 69
Table 3: Heliostat Sun spot sizes as measured from background subtracted CCD images taken at Sandia.
27
Timing Method Small Pulseheights Large Pulseheights All Data
σt σt σt
Conventional TDCs 0.82 ns 0.87 ns 0.86 ns
Waveform Digitizers 0.73 ns 0.51 ns 0.60 ns
Table 4: Comparison of measured timing resolution for conventional TDCs and 1 GSample/sec waveform
digitizers. The waveform digitizers have superior timing resolution at all pulseheights.
28
????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????
????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????
BBBBBBBBBBBBBBBBBBBBBBBBB
BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB
????????????????????????????????????????
BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB
BBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBBB
Gamma−ray
Cherenkov light poo l(200 meters diamete r)
heliostats
Solartower
Secondaryreflector
Camerabox withPMT’s
STACEE CONCEPT
Figure 2.1: The STACEE concept. Cherenkov light produced in the air shower created by an astrophysical γ-ray
is beamed to the ground. Solar heliostat mirrors reflect this light to a secondary collector on the central tower which
in turn reflects it to a camera of photomultiplier tubes.
29
Figure 2.2: STACEE telescope design, including secondary optic and camera. The STACEE instrument will
consist of three such telescope modules. One of these has already been constructed and tested successfully in the
field. Dimensions are given by the scale on the right.
30
Discriminator
Scaler
TDC
TRIGGER
ANALOG:
Combine PMTsignals infan−in, anddiscriminateon sum
DIGITAL:
Combine theindividuallydiscriminatedPMTs withmultiplicitylogic
PMTFilterandPre−Amp
CableDelay
WaveformDigitizer(Tektronix scopes)
LinearFan−Out
ADC
OtherPMTs
To ADC, TDC,digitizers
Figure 2.3: Schematic of the electronics setup used for the prototype tests (one channel of eight). Each PMT signal
was AC-coupled and amplified. Cable delays compensated for the varying times-of-flight between heliostats and the
central tower. Each signal was then split by a fan-out. Analog copies of the signal went to waveform digitizers and
ADC units. The signal was also discriminated, and the discriminated outputs drove scaler units and stoped TDCs.
The trigger was formed by combining signals from all eight PMTs, and the trigger gated the waveform digitizers
and ADCs, and started the TDC units.
31
-150 -100 -50 0 50 100 150East/West Position (m)
0
125
250
Nor
th/S
outh
Pos
itio
n (m
)
TOWER
Heliostat PositionSelected Heliostats for August TestSelected Heliostats for October Test
Figure 2.4: Layout of the Sandia heliostat field. Each square represents the location of a single heliostat. The
sizes of the squares are not proportional to the physical size of the heliostats. The two different eight-heliostat
configurations used in the August and October tests are shown.
32
Month
0
50
100
150
200
Ave
rage
Num
ber
of O
bser
vabl
e H
ours
/ M
onth
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
Figure 3.1: Average number of hours of clear, moonless observing time expected per month. The expected fraction
of clear skies is taken from daily meteorological records for rain fall and percent sunshine for Albuquerque from 1948
to the present.
33
Celestron 8" SCT f/10
PMT
UVB Filters
Guide Ocular
ScalerLap TopComputer
Figure 3.2: Schematic drawing of the photometer setup used at Sandia.
34
Figure 3.3: Response curves for the standard photometric UBV filters [20].
35
0 20 40 60 80
Time (sec)
103
104
105
106
Cou
nts
/ Bin
Background
Signal
Figure 3.4: Light curve for the star Mu Andromeda showing the PMT counting rate for a single drift scan using
the V-band filter. Time bins are 0.1sec.
36
0 1 2 3SEC(Zenith Angle)
104
105
Stel
lar
Flu
x (A
rbit
rary
Uni
ts)
DataFit to Data
Figure 3.5: Plot of stellar flux versus atmospheric depth for the star Mu Andromeda.
37
0 20 40 60 80Zenith Angle (degrees)
0.00
0.25
0.50
0.75
1.00
Atm
osph
eric
Tra
nsm
issi
on
Figure 3.6: Atmospheric transmission for the blue waveband, which contains most of the observable Cherenkov
light, as a function of zenith angle. The points and curves are identified in the legend.
38
0.0 2.0 4.0 6.0 8.0PMT id
0.0
2.00*108
4.00*108
6.00*108
8.00*108
Sing
le P
.E. R
ate
(Hz)
Night SkyEdge On
Figure 3.7: Single photoelectron rates for the camera PMTs for two heliostat conditions: edge facing the secondary,
and viewing the night sky. Note that the statistical errors on the data are smaller than the point sizes.
39
0 100 200 300 400 500 600Time from Start of Run (s)
0
200
400
600
PM
T C
urre
nt (
Arb
itra
ry U
nits
)
0 100 200 300 400 500 600Time from Start of Run (s)
0
200
400
600
PM
T C
urre
nt (
Arb
itra
ry U
nits
)
Figure 4.1: Currents in two PMTs during a drift scan of the star Aldebaran.
40
0 200 400 600 800 1000 1200Time from Start of Run (s)
220
270
320
PM
T C
urre
nt (
Arb
itra
ry U
nits
)
0 200 400 600 800 1000 1200Time from Start of Run (s)
220
270
320
PM
T C
urre
nt (
Arb
itra
ry U
nits
)
Figure 4.2: PMT currents for two channels recorded while the heliostats were tracking a bright star. Note the
suppressed zeroes on the vertical scales.
41
Figure 4.3: CCD image of a single heliostat Sun spot projected onto the tower. The contours represent the fraction
of the total light contained within a given radius, starting at 10% and increasing in steps of 20%. The box underneath
is a 1m× 1m box to indicate scale.
42
Figure 4.4: Solar spot size (FWHM) projected on a target at the tower versus distance between heliostat and target
for 15 heliostats measured using a CCD camera. Except for one heliostat with particularly poor optical alignment
(8E3 at 107m) the spots follow a regular trend, becoming less concentrated with greater distance from the tower.
43
250 350 450 550 650Wave Length (nm)
0.0
0.25
0.5
0.75
1.0
Ref
lect
ivit
y
Figure 4.5: Measured heliostat facet reflectivity, as a function of wavelength.
44
Figure 5.1: A CCD image of the Sandia heliostat field projected onto a white lucite placard at the focal plane
of the secondary used for the October 1996 observations. This image was obtained by viewing the placard at an
oblique angle from the balcony where the telescope mount is installed. Grid lines drawn on the placard represent a
1 cm spacing at the focal plane. Images of eight selected heliostats are distinct and well-separated, with virtually
no optical overlap.
45
Figure 5.2: Analog trigger rate as a function of discriminator threshold. The data are compared to a model of
single heliostat triggers from cosmic rays and muons, as identified in the legend. The heliostats were viewing separate
parts of the sky.
46
Figure 5.3: Digital trigger rate as a function of individual channel discriminator threshold. The heliostats were
viewing separate parts of the sky. The lines represent power law fits to the data, for different trigger configurations,
as identified in the legend.
47
1 101 102 103
PMT Pulse Height
101
102
103
Cou
nts
/ Bin
Figure 5.4: Pulse height spectrum obtained from observations of air showers at the zenith.
48
0.0 20.0 40.0 60.0 80.0Zenith Angle θ (degrees)
0.0
1.0
2.0
3.0
4.0
Tri
gger
Rat
e (H
z)
DataFit to cos2(θ)
Figure 5.5: Trigger rate as a function of zenith angle.
49
Shower Max
Air Shower
Heliostat
Field of View
Heliostats
Figure 5.6a: Diagram illustrating overlapping heliostat
fields-of-view when the heliostats are all aimed at a source
at infinity. Note that the entire track of the air shower is
not completely contained within the overlap of the helio-
stat fields-of-view. Angles and distances have been exag-
gerated for purposes of illustrating the concept.
Shower Max
Air Shower
Heliostat
Field of View
Heliostats
Figure 5.6b: Diagram illustrating the advantage of con-
vergent viewing. When the heliostats are canted in to
view the interaction region the air shower is better con-
tained by the overlapping heliostat fields-of-view.
50
-0.8 -0.4 0.0 0.4 0.8Canting Angle (degrees)
0.0
1.0
2.0
3.0
Tri
gger
Rat
e (H
z)
Data
Gaussian Fit
Figure 5.7: Trigger rates as a function of heliostat canting angle. A canting angle of 0◦ corresponds to parallel
viewing. A 4/8 digital trigger was used and the heliostats were viewing the zenith.
51
101 102 103
Energy (GeV)
10-2
10-1
1
101
102
Pho
ton
Den
sity
(γ/
m2 ) Gamma
Proton
Figure 5.8: The density of Cherenkov photons on the ground, within 100m of the core, for vertically incident γ-ray
and proton initiated air showers as determined by the Hillas Monte Carlo. The measured field of view of the Sandia
heliostats has been accounted for so that only those photons which would be collected by the STACEE secondary
telescope have been considered. Note that the statistical errors are smaller that the size of the symbols used in
plotting the data. 52